RU2666628C1 - Coding of samples array for low delay - Google Patents

Abstract

FIELD: data processing.SUBSTANCE: invention relates to means for encoding an array of samples. Entropy coding of the current part of the predetermined entropy slice is based not only on the corresponding estimates of the probability of a predetermined entropy slice, that adapt using the previously encoded part of the predetermined entropy slice, but also on the probability estimates, that are used for entropy coding of spatially adjacent, orderly entropy slices of the previous entropy slice in its neighboring part. Thereby estimate of probability used for entropy coding are adapted to the actual statistical information of the symbol more approximate, thus reducing the reduction in coding efficiency, usually caused by lower delay concepts. Time correlations are used additionally or alternatively.EFFECT: technical result is to reduce the delay in encoding the array of samples.20 cl, 4 tbl, 20 dwg

Description

This application relates to encoding an array of samples (sampling elements), such as encoding images or video.

Parallelization of the encoder and decoder is very important because of the increased processing requirements due to the HEVC standard, as well as due to the expected increase in video resolution. Multi-core architecture is becoming available in a wide range of modern electronic devices. Consequently, effective ways of enabling the use of a multi-core architecture are required.

LCU encoding or decoding occurs during raster scanning, with which the probability of CABAC is adapted to the specific features of each image. Spatial dependencies exist between adjacent LCUs. Each LCU (largest coding unit) depends on its left, top, top-left and top-right neighboring LCUs, due to various components, for example, motion vector, prediction, intra prediction and others. Because of the ability to parallelize during decoding, these dependencies usually need to be interrupted or interrupted in modern applications.

Some parallelization concepts have been proposed, namely wave data processing. The motivation for further research is the use of methods that reduce the loss of coding efficiency and thus reduce the cost of the bit stream for parallelization approaches in the encoder and decoder. In addition, low latency processing was not possible using the available methods.

Thus, it is an object of the present invention to provide a coding concept for arrays of samples that provides the possibility of less delay with a relatively less degradation in coding efficiency.

This task is provided by the contents of the attached independent claims.

If the entropy encoding of the current part of the predefined entropy slice is based on the corresponding probability estimates of the predefined entropy slice, which are adapted using the previously encoded part of the predefined entropy slice, but not only, but also on the probability estimates that are used in the entropy encoding of a spatially adjacent entropy in order of entropy slices of the previous entropy slice in its neighboring part, then the probability estimates used in entropy coding Manners adapt to the actual statistical information of the characters more closely, thereby reducing the reduction in coding efficiency typically caused by lower latency concepts. Temporary relationships can be used additionally or alternatively.

For example, the dependence on probability estimates that are used in the entropy coding of a spatially adjacent, in the order of entropy slices of a previous entropy slice, may involve initialization of probability estimates at the beginning of the entropy coding of a predetermined entropy slice. Typically, probability estimates are initialized to values adapted to the statistical information of the characters of the characteristic mixture of material in the sample array. To avoid transmitting the initialization values of the probability estimates, they are known to the encoder and decoder by definition. However, such predefined initialization values are naturally just a compromise between the bit rate (bit rate) of the additional information on the one hand and coding efficiency on the other hand, since the initialization values naturally - more or less - deviate from the actual typical statistical information of the currently encoded array material samples. Adaptation of probability during encoding of the entropy slice adapts the probability estimates to the actual statistical information of the symbols. This process is accelerated by initializing the probability estimates at the beginning of the entropy coding of the current / predetermined entropy slice, using the already adapted probability estimates of the just mentioned spatially neighboring, in the order of the entropy slices of the previous entropy slice, since the latter values are already, to some extent, adapted to the actual statistical information of the symbol of the sample array currently being examined. Low-latency coding can, however, be available by using probability estimates for initialized / current entropy slices when initializing, probability estimates are used in its neighboring part, instead of being found at the end of the entropy coding of the previous entropy slice. Using these measures, wave data processing is still possible.

Additionally, the aforementioned dependence on the probability estimates that are used in entropy coding of a spatially adjacent, in the order of entropy slices of the previous entropy slice, may involve an adaptation process consisting of adapting probability estimates directly used in entropy coding of the current / predetermined entropy slice. Adaptation of the probability estimate involves the use of the just-encoded part, that is, the just-encoded character (s), to adapt the current state of the probability estimates to the actual statistical information of the symbols. Using these measures, initialized probability estimates are adapted with some speed of adaptation to the actual statistical information of the symbol. This adaptation speed is increased by performing the just mentioned adaptation of the probability estimate, based not only on the currently encoded symbol of the current / predetermined entropy slice, but also depending on the probability estimates that were found in the neighboring part of the spatially neighboring, in the order of the entropy slices of the previous entropy slice. Again, by selecting the spatially adjacent current part of the current entropy slice and the neighboring part of the previous entropy slice, respectively, wave data processing is still possible. The advantage of combining the adaptation of one’s own probability estimate for the current entropy slice with the adaptation of the probability of the previous entropy slice is the increase in the rate at which the adaptation to the actual statistical information of the symbols occurs, since the many symbols through which the current and previous entropy slices move through contribute to the adaptation, not just the symbols of the current entropy slice.

Preferred embodiments of the embodiments of the present invention are the subject of the dependent claims. Additionally, preferred embodiments are described with respect to the figures, among which:

FIG. 1 shows a block diagram of an example encoder;

FIG. 2 shows a schematic diagram of dividing an image into slices and parts of slices (i.e. blocks or coding units) along with the coding order defined between them;

FIG. 3 shows a flowchart of an exemplary encoder, such as the encoder of FIG. one;

FIG. 4 shows a schematic diagram for explaining the operation of an example encoder, such as the encoder in FIG. one;

FIG. 5 shows a schematic diagram of a parallel working embodiment of an encoder and a decoder;

FIG. 6 shows a block diagram of an example decoder;

FIG. 7 shows a flowchart of an example decoder, such as the decoder of FIG. 6;

FIG. 8 shows a schematic diagram of an example bitstream resulting from the coding scheme of FIG. 16;

FIG. 9 schematically shows an example of calculating probability using other LCUs;

FIG. 10 shows a graph illustrating the results of RD for the internal method (4 streams) compared to HM3.0;

FIG. 11 is a graph illustrating RD results for a low latency method (1 stream) compared to HM3.0;

FIG. 12 is a graph illustrating RD results for a random access method (1 stream) compared to HM3.0;

FIG. 13 is a graph illustrating RD results for a low latency method (4 streams) compared to HM3.0;

FIG. 14 schematically and approximately illustrates possible entropy slices compounds;

FIG. 15 schematically and approximately illustrates a possible signaling of an entropy slice;

FIG. 16 schematically and approximately illustrates the encoding, segmentation, interleaving, and decoding of entropy slice data through fragments;

FIG. 17 schematically and approximately illustrates a possible connection between frames;

FIG. 18 schematically and roughly illustrates the possible use of related information;

FIG. 19 schematically shows the possibility of a wave propagating obliquely within a spatial / temporal space filled with successive arrays of samples; and

FIG. 20 schematically shows another example for subdividing entropy slices into fragments.

To facilitate understanding of the following measures to improve low latency with fewer disadvantages with respect to coding efficiency, the encoder in FIG. 1 is first described more generally without first discussing preferred concepts of embodiments of the present application and how it can be integrated into the embodiment of FIG. 1. It should be mentioned, however, that the structure shown in FIG. 1 serves merely as an illustrative environment in which embodiments of the present application may be used. Generalizations and alternatives for encoders and decoders in accordance with embodiments of the present invention are also briefly discussed.

FIG. 1 shows an encoder for encoding an array of 10 samples into an entropy encoded data stream 20. As shown in FIG. 1, the sample array 10 may be one of the sequences 30 of sample arrays, and the encoder may be configured to encode the sequence 30 into the data stream 20.

The encoder in FIG. 1 is generally indicated by a reference symbol 40 and comprises a pre-encoder 42, followed by an entropy encoding stage 44, the output of which is output from the data stream 20. The precoder 42 is configured to receive and process an array of 10 samples to describe its contents through syntax elements with a predetermined syntax, each syntax element being a corresponding one of a predetermined set of types of syntax elements, which, in turn, are associated with corresponding semantics.

When describing an array of 10 samples using syntax elements, pre-encoder 42 may subdivide an array of 10 samples into coding units 50. The term “coding unit” may, for reasons indicated in more detail below, be alternatively referred to as “coding tree units” (CTUs). One possibility of how pre-encoder 42 can subdivide an array of 10 samples into coding units 50 is shown by way of example in FIG. 2. In accordance with this example, this division regularly divides the array of 10 samples into coding units 50, so that the latter are arranged in rows and columns, so that without a gap cover the entire array of 10 samples without overlap. In other words, pre-encoder 42 may be configured to describe each coding unit 50 by means of syntax elements. Some of these syntax elements may generate unit information for an additional unit of the corresponding coding unit 50. For example, by dividing into multiple trees, dividing information may describe the division of the corresponding coding unit 50 into prediction blocks 52, with the pre-encoder 42 associating the prediction mode with the associated prediction parameters for each of these prediction blocks 52. This prediction unit may allow prediction blocks 52 to be of different sizes, as illustrated in FIG. 2. The preliminary encoder 42 may also associate the residual subdivision information with the prediction blocks 52 to further subdivide the prediction blocks 52 into the residual blocks 54 to describe the prediction error for each prediction block 52. Thus, the precoder can be configured to generate a syntax description of the array of 10 samples in accordance with the hybrid coding scheme. However, as noted above, the just-mentioned method by which the pre-encoder 42 describes an array of 10 samples by means of syntax elements was presented merely for purposes of illustration, and it may also be embodied differently.

The precoder 42 may use spatial relationships between the contents of neighboring coding units 50 from an array of 10 samples. For example, precoder 42 may predict syntax elements for a certain coding unit 50 from syntax elements defined for previously encoded coding units 50 that are spatially adjacent to the currently encoded coding unit 50. In FIG. 1 and 2, for example, the upper and left neighboring elements serve for prediction, which is illustrated by arrows 60 and 62. In addition, the preliminary encoder 42 in the intra prediction method can already extrapolate the encoded contents of neighboring coding units 50 into the current coding unit 50 to obtain sample prediction current coding unit 50. As illustrated in FIG. 1, precoder 42 may, without spatial relationships, in the time domain, predict samples and / or syntax elements for the current coding unit 50 from previously encoded sample arrays, as illustratively shown in FIG. 1 by arrow 64. That is, this motion-compensated prediction can be used by the pre-encoder 42, and the motion vectors themselves can be a temporal prediction object from the motion vectors of previously encoded sample arrays.

That is, pre-encoder 42 may describe the contents of an array of 10 samples with respect to coding units, and it may use spatial prediction for this purpose. Spatial prediction is limited for each coding unit 50 to spatially adjacent coding units of the same array of 10 samples such that when coding order 66 is kept among the coding units 50 of the array of 10 samples, through adjacent coding units serving as a prediction support for spatial prediction, in general, coding order 66 has already been moved before the current coding unit 50. As shown in FIG. 2, the coding order 66 defined among coding units 50 may, for example, be a raster scan order according to which coding units 50 move line by line from top to bottom. Optionally, subdividing array 10 into an array of tiles can cause the scan order 66 to move — in raster scan order — through the coding units 50 that create one tile, first before moving on to the next tile in order, which, in turn, can also have a raster scan type. For example, spatial prediction may involve simply adjacent coding units 50 within a line of coding units above a line of coding units within which the current coding unit 50 is located, and a coding unit within the same line of coding units, but to the left of the current coding unit. As will be explained in more detail below, this restriction on spatial relationship / spatial prediction provides the possibility of parallel wave data processing.

The precoder 42 directs the syntax elements to the entropy coding stage 44. As just stated, some of these syntax elements are prediction encoded, i.e. represent prediction errors. The precoder 42 may thus be regarded as a predictive encoder. In addition, the precoder 42 may be a transform encoder configured to transform the remainder of the prediction codes of the contents of the encoding units 50.

An exemplary internal structure of entropy coding stage 44 is also shown in FIG. 1. As shown, the entropy coding stage 44 may optionally comprise symbolic representation for converting each of the syntax elements received from the preliminary encoder 42, the number of possible states of which exceeds the number of elements of the alphabet of characters, into a sequence of characters s _i of the character alphabet based on which the means (engine) 44 of entropy coding works. In addition to this optional symbolic representation means 70, entropy encoding means 44 may include context selection means 72 and initialization means 74, probability estimation control means 76, probability estimation adaptation means 78 and entropy encoding core 80. The output of the entropy coding core generates the output of the entropy coding stage 44. In addition, entropy encoding core 80 contains two inputs, namely, one for receiving symbols s _{i of a} sequence of symbols, and the other for receiving estimates of probability p _i for each of the symbols.

Due to the properties of entropy coding, the coding efficiency with respect to the compression ratio increases with an improvement in the probability estimate: the better the probability estimate corresponds to the actual statistical information of the symbol, the better the compression ratio.

In the example of FIG. 1, the context selector 72 is configured to select, for each character s _{i, a} corresponding context c _i from a set of available contexts that are controlled by the control means 76. It should be noted, however, that the choice of context forms a merely optional feature and can not be used, for example, using the same context for each character. However, if context selection is used, then context selection tool 72 may be configured to perform context selection, at least in part based on information that relates to coding units outside the current coding unit, namely, which refers to neighboring coding units within a limited proximity, as discussed above.

The control means 76 comprises a storage device that stores, for each available context, a probability estimate associated therewith. For example, a character alphabet may be a binary alphabet, so just one probability value will probably have to be stored for each available context.

Initialization tool 74 may periodically initialize or reinitialize probability estimates stored in control means 76 for available contexts. Possible points in time at which such initialization can be performed are discussed further below.

The adaptation means 78 has access to symbol pairs s _i and the corresponding probability estimates p _i and accordingly adapts the probability estimates in the control means 76. That is, each time a probability estimate is applied by the entropy encoding core 80 to entropy encode the corresponding symbol s _i into the data stream 20, the adaptation means 78 can change this probability estimate in accordance with the value of the current symbol s _i so that this estimate p _i The probability was better adapted to the actual statistical information of the symbol when the next symbol that is associated with a given probability estimate (through its context) is encoded. That is, the adaptation means 78 receives a probability estimate for the selected context from the control means 76 along with the corresponding symbol s _i and adapts the probability estimate p _i , respectively, so that an adaptive probability estimate is used for the next symbol s _{i of} the same context c _i .

The entropy coding core 80, for example, is configured to operate in accordance with an arithmetic coding scheme or with an entropy coding scheme with a separation of a probability interval. In arithmetic coding, the entropy coding core 80, for example, continuously updates its state when it encodes a sequence of characters, with a state, for example, defined by a probability interval defined using a probability interval width value and a probability interval offset value. When operating in the pipeline concept, the entropy coding core 80, for example, divides the range of possible values of the probability estimates into different intervals with performing entropy coding with a fixed probability with respect to each of these intervals, thereby obtaining a substream for each of the sub-intervals whose coding efficiency, accordingly, it is configured for the associated probability interval. In the case of entropy encoding, the output of the data stream 20 is an arithmetically encoded data stream that transmits information to the decoding side, which provides the ability to emulate or re-execute the process of dividing this interval.

Naturally, it is possible for entropy coding state 44 to entropy encode all the information, that is, all syntactic elements / symbols s _i belonging to an array of 10 samples, by initializing probability estimates just once at its beginning, and then continuously updating the probability estimate using 78 adaptations. This, however, would lead to the creation of a data stream 20 that would have to be sequentially decoded on the decoding side. In other words, it would not be possible for any decoder to subdivide the resulting data stream into several sub-portions and decode the sub-portions in parallel. This, in turn, would prevent all efforts to reduce the delay.

Accordingly, as will be described in more detail below, it is convenient to subdivide the data set describing an array of 10 samples into so-called entropy slices. Each of these entropy slices, respectively, covers a different set of syntax elements related to an array of 10 samples. If cascade 44 of entropy encoding, however, entropy encoded each entropy slice completely independently of each other, first initializing the probability estimate once and then constantly updating the probability estimates separately for each entropy slice, then the encoding efficiency would decrease due to an increased percentage of the data that relate and describe an array of 10 samples for which the used probability estimates are (nevertheless) less accurately adapted to the actual statistical information symbol but.

In order to overcome the problems just mentioned while matching the need for a low coding delay on the one hand and high coding efficiency on the other, the following coding scheme can be used, which is described below with respect to FIG. 3.

First, the data describing the array of 10 samples is divided into portions, called hereinafter “entropy slices”. Unit 80 does not have to be without overlays. On the other hand, this division may at least partially correspond to the spatial division of the array of 10 samples into different portions. That is, according to subdivision 80, the syntax elements describing the array 10 of samples can be allocated to different entropy slices depending on the location of the coding unit 50, which refers to the corresponding syntax element. See, for example, FIG. 2. FIG. 2 shows an example subdivision of an array of 10 samples into different portions 12. Each portion corresponds to a corresponding entropy slice. As an example, it is shown that each portion 12 corresponds to a line of coding units 50. Other divisions, however, may also be performed. However, it is preferable if the division of the array of samples 10 into portions 12 adheres to the aforementioned encoding order 66, so that the portions 12 cover consecutive passes of the encoding units 12 in encoding order 66. Even so, however, it is not required that the start and end positions of the portion 12 in coding order 66 coincide with the left and right edges of the lines of coding units 50, respectively. It is not even necessary to match the boundaries of the coding units 50, which immediately follow one another, and the coding order 66.

Using this division of the array of 10 samples among the portions 12, the order of 16 entropy slices is determined, according to which the portions 12 follow each other in coding order 66. In addition, for each entropy slice, a corresponding entropy encoding path 14 is determined, namely, a fragment of the encoding path 66 directed to the corresponding portion 12. In the example of FIG. 2, where the portions 12 coincide with the lines of the coding units 50, the entropy encoding paths 14 from each point of the entropy slice along the line direction are parallel to each other, that is, in this case, from the left side to the right side.

It should be noted that it would be possible to limit the spatial predictions performed by the precoder 42 and the context obtaining performed by the context selector 72 so that they do not cross the slice boundaries, that is, so that the spatial predictions and the context selection results are independent of the data corresponding to another entropy slice. Thus, “entropy slices” correspond to the usual definition of “slices” in H.264, for example, which are decoded completely independently from each other, with the exception of the probability initialization / adaptation dependence below. However, it is also acceptable to provide the possibility that spatial predictions and context selection results, i.e. generally speaking dependencies, cross slice boundaries to use local / spatial interdependencies, since WPP processing can still be performed while precoding inversion is considered, then there is a restoration based on syntactic elements, and a choice of the entropy context. Until now, the entropy slices have somehow corresponded to “dependent slices”.

Unit 80 may, for example, be performed using entropy coding stage 44. The division may be fixed or may vary among multiple sequences 30. The division may be established by definition or may be communicated in the data stream 20.

Based on the entropy slices, actual entropy coding may occur, that is 82. For each entropy slice, the entropy coding may be structured into an initial phase 84 and a continuation phase 86. The initial phase 84 involves, for example, initializing probability estimates, as well as invoking the actual entropy coding process for the corresponding entropy slice. The actual entropy coding is then performed during the continuation phase 86. Entropy encoding during phase 86 is performed along the corresponding entropy encoding path 14. The initial phase 84 for each entropy slice is controlled in such a way that the entropy coding of multiple entropy slices is started sequentially using an order of 16 entropy slices.

Then, in order to avoid the above drawback, which is the result of the entropy encoding of each entropy slice completely independently of each other, the entropy encoding process 82 is controlled so that the current part, for example, the current encoding unit of the current entropy slice, is entropy encoded based on the corresponding probability estimates of the current entropy slice, which adapt using previously encoded parts of the current entropy slice, that is, parts of the current en of the tropical slice to the left of the current coding unit 50 in the case of FIG. 2, and on the probability estimates that are used in the entropy coding of spatially adjacent, in order of 16 entropy slices of previous entropy slices in its neighboring part, that is, in the neighboring coding unit.

To describe the above relationship more clearly, reference is made to FIG. 4. FIG. 4 illustrates the nth, nth, and n + 1th entropy slices in order of 16 entropy slices with a reference character 90. Each entropy slice 90 covers a sequence of syntax elements describing a portion 12 of an array of 10 samples to which a corresponding entropy slice 90. Along the path 14 of entropy encoding, the entropy slice 19 is segmented into a sequence of segments 92, each of which corresponds to one of the encoding units 50 of portion 12, to which the entropy slice 90 belongs.

As described above, the probability estimates used in entropy encoding of entropy slices 90 are continuously updated during phase 86 of continuing along entropy encoding path 14 so that the probability estimates are better and better adapted to the actual statistical information of the symbol of the corresponding entropy slice 90 - i.e., estimates probabilities are associated with the corresponding entropy slice. Although the probability estimates 94 used to entropy encode the entropy slice 90, continuation is continuously updated during phase 86, in FIG. 4 simply states of probability estimates 94 that appear in the start and end positions of segments 92 are illustrated and are further mentioned below. In particular, the state before entropy encoding of the first segment 92, which is initialized during the initial phase 84, is shown as 96, the state detected after encoding the first fragment is illustrated as 98, and the state detected after encoding the first two fragments is indicated as 100. Those the same elements are shown in FIG. 4 also for the entropy slice n-1 in order of 16 entropy slices, and the next entropy slice, i.e., the entropy slice n + 1.

Now, to ensure the above dependency, the initial state 96 for entropy encoding of the nth entropy slice 90 is set depending on any intermediate state of probability estimates 94 detected during encoding of the previous entropy slice n-1. An “intermediate state” shall mean any state of probability estimates 94, excluding the initial state 96 and the final state detected after entropy encoding of the entire entropy slice n-1. By doing so, the entropy coding of the sequence of entropy slices 90 in order of 16 entropy slices can be parallelized with the degree of parallelization, which is determined by the ratio of the number of segments 92 preceding the state used to initialize the probability estimates 94 for entropy encoding of the next entropy slice, i.e., a, and the number of segments 92 following this stage, i.e. b. In particular, in FIG. 4a, for example, is set equal to, by initialization, that is, adaptation, state 100, so as to set state 96 of the current entropy slice to state 100 of the previous entropy slice, illustrated by arrow 104.

With these measures, the entropy coding of any segment 92 after state 100 in order 14 of the trajectory of the entropy coding will depend on the probability estimate 94, which is adapted during the continuation phase 86, based on previous fragments of the same entropy slice, as well as on the probability estimate, which used in entropy coding of the third segment 92 of the previous entropy slice 90.

Accordingly, entropy encoding of entropy slices 90 may be performed in parallel with pipeline scheduling. The only restrictions introduced during planning are that the entropy coding of a certain entropy slice can begin immediately after the end of the entropy coding of the a-th segment 92 of the previous entropy slice. The entropy slices 90 immediately after each other in order of 16 entropy slices are not subject to any other restrictions regarding the time alignment of the entropy coding procedure during the continuation phase 86.

However, in accordance with another embodiment, a more intense compound is additionally and / or alternatively used. In particular, as illustrated in FIG. 4 using the usual arrows 106, adapting the probability estimate during the continuation phase 86 causes the data of the coding unit corresponding to some segment 92 to change the probability estimates 94 from the state at the beginning of the corresponding segment 92 to the end of this segment 92, thereby improving the approximation actual statistical information of the symbol as indicated above. That is, for the entropy slice n-1, adaptation 106 is performed that simply depends on the data of the entropy slice n-1, and the same applies to adaptation 106 of the probability estimate for the entropy slice n, etc. For example, you can perform initialization, as explained above using arrows 104, by performing adaptation 106 of the probability estimate without further mutual influence between the entropy slices 90. However, to speed up the approximation of the probability estimate to the actual statistical information of the symbol, adapting 106 the probability estimate of successive entropy slices can be connected so that adaptation 106 of the probability estimate of the previous n-1 entropy slice also influences or is taken into account when the cut adapts tat adapting probability estimates current entropy slice n. This is illustrated in FIG. 4, using arrow 108 facing away from state 110 of spatially adjacent probability estimates 94 for entropy encoding of the nth entropy slice 90 to state 100 of estimates 94 of the probability of entropy coding of the nth entropy slice 90. When using the above initialization of state 96, compound 108 probability adaptation can, for example, be used in any of the b probability estimation states that are found after entropy coding of b segments 92 of the previous entropy slice. More precisely, the probability estimates that are detected immediately after the entropy coding of the first segment 92 of the current entropy slice can be the result of the usual adaptation of 106 probability and by taking into account 108 states of the probability estimates resulting from the adaptation of 106 probability estimates during the entropy coding of (a + 1) segment 92 of the previous entropy slice n-1. “Accounting” may, for example, involve some averaging operation. An example will be further described below. In other words, state 98 of the probability estimates 94 of the entropy encoding of the nth entropy slice 90 at the beginning of the entropy encoding of its segment 92 may be the result of averaging the previous state 96 of the probability estimates 94 for entropy encoding of the current entropy slice n, which are adapted using adaptation 106, and the state before the entropy encoding of the (a + 1) th segment 92 of the previous entropy slice n-1, modified according to probability adaptation 106. Similarly, state 100 may be the result of averaging the result of adaptation 106 performed during entropy coding of the current entropy slice n, and the result of adapting the probability during entropy coding of the (a + 2) th segment 92 of the previous entropy slice n-1, etc.

More specifically, it is believed that

p (n) -> {i, j}, where i, j denotes the position of any coding unit (moreover, (0,0) is the upper left position and (I, J) is the lower right position), i∈ {1 .. .I} and j∈ {1 ... J}, I is the number of columns, J is the number of rows and p () determines the order of 66 trajectories,

P _{{i, j}} is the probability estimate used in entropy coding of the coding unit {i, j}; and

T (P _{{i, j}} ) is the result of adaptation 106 of the probability P _{{i, j}} based on the coding unit {i, j};

Then, the probability estimates 106 of successive entropy slices 90 can be combined to replace the usual internal adaptation of the entropy slice according to P _{p (n + 1)} = T (P _{p (n)} ) by

P _{p (n + 1)} = average value (T (P _{p (n)} ), T (P _{{i, j} 1} ), ..., T (P _{{i, j} N} )),

where N can be 1 or more than 1 and {i, j} _{1 ... N is} chosen from (are within) any previous (in order of 16 entropy slices) entropy slice 90 and associated portion 12, respectively. The “mean value” function may be one of a weighted summation, a median function, etc. p (n) = {i, j} is the current coding unit, and p (n + 1) is next in accordance with coding orders 14 and 66, respectively. In the presented embodiments, p (n + 1) = {i + 1, j}.

Preferably, {i, j} _{1 ... N} corresponds for each k∈ {1 ... N} {i, j} _{1 ... N} = {i _k , j _k } and i _k <i + 3 and j _k <j, and p (n) = {i, j} is the current coding unit (that is, any of the second or subsequent coding units of the current entropy slice), that is, they are not located outside the wave.

In the latter alternative embodiment, the time planning of entropy coding, when entropy coding of entropy slices 90 is performed in parallel, is more intensively connected to each other. That is, during the continuation phase 86, the next segment 92 on the line of the current entropy slice may begin immediately after the end of the corresponding segment of the previous entropy slice, which is located further in the order of the entropy coding path 14.

In other words, the above discussion shows an example in which the decoder 40, and in particular the entropy encoding stage 44, is configured to perform for the entropy slice 90, such as the nth, to initialize its probability estimates 94 before decoding the first segment 92 corresponding to the first coding unit / to the first block 50 of portion 12, corresponding to the nth entropy slice along the corresponding encoding path 14, using probability estimates found after entropy decoding of the second unit gical encoding / block 50 of portions 12, 16 corresponding to the order entropy slice, the previous entropy slice along the respective trajectories 14 encoding. Additionally, or alternatively, the decoder 40 and, in particular, the entropy encoding stage 44 can be configured to perform entropy decoding for each entropy slice 90 and adapt the probability estimates in such a way that, after the current encoding part / block / unit 50 of the current the entropy slice was entropy decoded based on the corresponding probability estimates 94 from the current entropy slice 90, the corresponding probability estimates 94 from the current entropy slice adapt to depending on this current part of the current entropy slice and probability estimates that are detected by entropy decoding of a neighboring encoding part / block / unit 50 of a spatially adjacent previous entropy slice, such as a slice in the aforementioned adjacent row in the second column to the right of the current part / block / coding unit current entropy slice.

As it became clear from the above discussion, the probability estimate must be adapted / managed for each entropy slice 50 separately. This can be done by sequentially processing the entropy slices and storing the states of the probability estimates as states that are shown and mentioned by way of example with respect to FIG. 4, i.e., 96, 98, 100, 110, and 102, on the corresponding probability estimation memory 120 (see FIG. 1). Alternatively, more than one entropy encoding stage 44 may be provided for decoding entropy slices in parallel. This is illustrated in FIG. 5, where a plurality of implementations of entropy encoding cascades 44 are shown, each is associated with a corresponding one of the entropy slices and corresponding portions 12 of an array of 10 samples, respectively. FIG. 5 also illustrates the decoding process and its possible implementations using parallel implementations of the corresponding entropy decoding stage 130. For each stage 130 of entropy decoding, a corresponding one of the entropy slices is transmitted through the data stream 20.

FIG. 5 shows that entropy coding stages 44 and corresponding decoding stages 130 do not work in parallel completely independently of each other. Instead, stored probability estimation states, such as those stored in memory 120, are transmitted from one stage corresponding to the corresponding entropy slice to another stage related to the entropy slice following in accordance with the order of 16 entropy slices.

FIG. 5 shows, for illustrative purposes, also a possible movement order for moving through the results of a possible separation of coding units 50, such as a moving order 140 defined among prediction units 52 within a single coding unit 50. For all of these prediction blocks 52, the corresponding syntax elements are contained within the corresponding segment 92 and correspond to the values of these syntax elements, probability estimates 94 are adapted while moving along the path 140, and adaptation while moving through coding units 50 is determined by the aforementioned “T”. In CABAC according to H.264 and HEVC, “T” is performed based on a table, that is, a “route table” that defines transitions from the current state of the probability estimate for the current context to the next state depending on the current symbol value associated with this context.

Before proceeding to FIG. 6, which shows an embodiment of a decoder corresponding to an embodiment of the encoder of FIG. 1, it should be noted that the predictive nature of the preliminary encoder 42 serves merely as an illustrative embodiment. In accordance with alternative embodiments, the pre-encoder 42 may be left with syntax elements that the entropy coding stage 44 operates with, which are, for example, the original sample values of the array of 10 samples. Even alternatively, pre-encoder 42 may be configured to decompose the subbands of the array of 10 samples, as in JPEG. An additional entity of the context selector 72 has already been mentioned above. The same applies with respect to the initialization means 74. The same may be implemented differently.

FIG. 6 shows a decoder 200 corresponding to encoder 40 in FIG. 1. From FIG. Figure 6 shows that the design of decoder 200 essentially reflects the design of encoder 40. That is, decoder 200 comprises an input 202 for receiving a data stream 20, followed by a series connection of the entropy decoding stage 204 and creation means 206. The entropy decoding stage 204 entropy decodes the entropy slices transmitted in the data stream 20 and sends, in turn, the decoded symbols s_i and syntax elements, respectively, to the creation tool 206, which in turn requests the syntax elements from the entropy decoding stage 204 using the corresponding request 208. In other words, the creation tool 206 is also responsible for analyzing the stream of syntax elements created by the preliminary encoder 42 in the limits of the encoder. Accordingly, the authoring tool 206 sequentially requests syntax elements from the entropy decoding stage 204. The entropy decoding stage 204 has essentially the same structure as the entropy encoding stage 44. Accordingly, the same reference signs of the internal blocks of the entropy decoding stage 204 are again used. The symbolic means 70, if present, converts the syntax elements requests into symbol requests, and the entropy decoding core 80 responds with the corresponding symbol value s_i, the means 70 representations in symbolic form displays the sequence of received characters forming the actual words of the characters with syntactic elements, and directs them to the means 206 creation. Creation means 206 restores an array 10 of samples from a stream of syntax elements received from entropy decoding stage 204, such as described above, using prediction decoding, etc. More specifically, the creation tool 206 also uses the encoding order 66 and performs decoding on the coding units to perform predictions 60, 62, and 64. One or more predictions for the syntax elements or for sample values are combined, for example, summed, if desired, using the remainder of the prediction derived from the syntax elements of a stream of syntax elements. The entropy decoding core 80, as well as the entropy coding core 80 supports the arithmetic decoding concept or the entropy decoding concept with a separation of the probability interval. In the case of arithmetic decoding, entropy decoding core 80 may continuously update the internal state in the form of a partial value of the interval width and such a value as an offset value indicative of this partial interval. The update is performed using an inbound data stream. The current partial interval is subdivided similarly to the entropy coding core 80 using the estimate p_i probabilities provided for each character s_i by means of a context selector 72 along with a probability estimation control means 76. The adaptation means 78 performs the adaptation of the probability estimate using the value s_i decoded character to update p values_i context probability estimates c_i, associated with the character s_iusing the context selector 72. Initialization using initialization means 74 is performed in the same implementations and in the same manner as on the encoding side.

The decoder in FIG. 6 operates similarly to the operation of the encoder as described above with respect to FIG. 3. At step 230, an array of 10 samples is divided into entropy slices. See, for example, FIG. 8. FIG. 8 shows a data stream 20 that arrives at input 202 and, in particular, entropy slices 90 contained therein. At step 230, each of these entropy slices 90 is associated with a portion 12, this corresponding entropy slice 90 is connected so that the above-described initialization of the probability estimate and adaptation of the probability estimate based on the corresponding previous entropy slice can be performed. The division of the array of samples, or more precisely, the binding of the entropy slices 90 with their corresponding portions 12, can be performed using the creation tool 206. Linking can be achieved through various measures, such as side information contained in the data stream 20 in non-entropy encoded portions, or by definition.

The entropy slices 90 are then, in the process of entropy decoding 232, entropy decoded in a manner that reflects the encoding processing 82, namely, for each entropy slice 90, performing an initial phase 234 and a continuation phase 236 with initialization of the probability estimate and adaptation in the same way and in the same implementations, as in the encoding procedure.

The same parallelization as described above with respect to encoding is possible on the decoding side. The implementations 130 of the entropy decoding stage shown in FIG. 5 may each be embodied, as shown with respect to the entropy decoding stage 204 in FIG. 6. A probability estimator 240 may be used to store the state of the probability estimates for use in the entropy decoding stage 130 responsible for entropy decoding of a subsequent entropy slice in entropy encoding order 16.

After describing embodiments of the present application, the concepts discussed are described below again using a different formulation. Several further aspects of the present application are described below. In the following, the aforementioned coding units 50 are called LCU (largest coding unit), thus adapting the wording to the promising HEVC standard.

First, the probability adaptation 106 discussed above is briefly discussed again with respect to FIG. 9.

The current LCU uses probability estimates, such as CABAC probabilities, available after encoding the previous left LCU. For example, the LCU in FIG. 9, denoted by x, is assumed to be entropy encoded using a probability estimate p ₁ that is adapted before the entropy coding of the left-handed LCUs to the left of the LCU x is completed. However, if the information is used not only from the left, but also from one or more other LCUs that are already processed and available, then better results can be provided when adapting the probability.

As described above, in entropy decoding and entropy encoding of entropy slices, new probability estimates are computed before encoding or decoding each LCU using existing probabilities (probability estimates) from other LCUs. More precisely, probability adaptation is performed not only from any previous LCUs of the current entropy slice, but also from LCUs of entropy slices preceding the order of entropy slices. This approach is again presented in FIG. 9. The initial probability of the current LCU, which is denoted by X in FIG. 9 can be calculated accordingly:

where a ₁ , ..., a _k are the LCU weighting factors.

It is tested which probability weighting provides the best results. In this experiment, only neighboring LCUs were used. The study shows the use of weighting: 75% of the left LCU and 25% of the upper right LCU. In FIG. 10-13 show the results. Charts entitled using prob. Adapt. ”, Use the probability adaptation described above.

However, when adapting a probability estimate, not only adjacent blocks can be used. Each nearest LCU has its own neighbors, the use of which to optimize the probability can be significant. In other words, not only the LCUs from the nearest top row can be applied. In FIG. 9 you can see an example in which neighbors were taken as the sources of probability estimates and each upper right LCU of each next upper row can be taken, compare p ₅ and p _k .

It must be recognized that some complexity was introduced using the foregoing recalculation of probability or the adaptation of a probability estimate. A new calculation of the probability estimate occurs, for example, in three stages: first, the probability estimates of each candidate must be obtained from each state of the context. This is done by saving to the storage device 120 and 240, respectively, or by directing the parallel processing of decoding the entropy slice n in such a way that these states are available simultaneously. Then, using equation (1), an optimized probability (p _new ) will be created. That is, some averaging, for example, can be used to combine adapted probability estimates of various entropy slices. As a last step, the new context state is converted from p _new and replaces the old one. That is, the probability estimation control means 76 borrows new probability estimates obtained in this way. This implementation for each syntax element, in particular, using the operations of multiplication, can extremely increase the complexity. The only way to eliminate this drawback is to try to avoid these three steps. If the number of candidates and their weights are determined, then a pre-calculated table can be approximated for each situation. Thus, only one simple access to the table data is needed using candidate indices (context states).

It is argued that this technique can provide good results for both applications - with and without entropy slices. The first application uses only one slice per frame, so the probability adaptation is optimized without any other changes. In the case of entropy slices, the adaptation of probabilities occurs within each slice, independently of the other slices. This allows for a quick study of the probabilities of the current LCU.

The above description also describes the use of a second upper line LCU, that is, the use of a second LCU to initiate a probability estimate. Parallelization of encoding and decoding is possible if certain conditions of the bit stream, which are mentioned above (entropy slices), are provided. The CABAC probability relationship between LCUs must be eliminated. Wave parallel processing is important to make the first LCU of each line independent of the last LCU of the previous line. This can be achieved if, for example, the CABAC probabilities are reinitialized at the beginning of each LCU line. However, this method is not optimal, because each reinitialization loses the provided CABAC probabilities, which are adapted to the specific features of the image. This disadvantage can be reduced if the initialization of the CABAC probabilities of the first LCU of each line occurs using the probabilities obtained after the second LCU of the previous line.

As described above, an increase in the probability adaptation speed can be achieved by connecting the probability adaptation of spatially adjacent entropy slices. In particular, in other words, the above discussion also anticipates a decoder, such as the decoder in FIG. 6, for restoring an array (10) of samples from an entropy encoded data stream, configured for entropy decoding (performed by the cascade of entropy decoding) of a multitude of entropy slices in a data stream of an entropy encoder to restore various portions (12) of an array of samples associated with entropy sequences, accordingly, by performing entropy decoding for each entropy slice along the corresponding entropy coding path (14), using corresponding probability estimates, adapting (performed using adaptation tool 78) corresponding probability estimates along the corresponding path of entropy coding, using the previously decoded portion of the corresponding entropy slice, triggering the entropy decoding of the set of entropy slices sequentially, using the order of (16) entropy slices, and performing at the entropy decoding a predetermined entropy slice entropy decoding of the current part (x) is predetermined entropy slice, based on the corresponding probability estimates of the predetermined entropy slice, which are adapted using the previously decoded part of the predetermined entropy slice (including p1, for example), and probability estimates, which are used in the entropy decoding of a spatially adjacent entropy slice of the previous entropy a slice (a slice containing X, for example) in a neighboring part (such as p4) of a spatially adjacent entropy slice.

Different portions may be block rows (for example, LCUs or macroblocks) of an array of samples. The latter may be a video. The trajectory of entropy coding can propagate along the lines. Entropy coding, and thus also adaptation of probability, can adapt to the context. The order of entropy slices can generally be chosen so that, in order of entropy slices, different portions follow each other in a direction (16) that is rotated relative to the paths (14) of entropy encoding of entropy slices, which, in turn, propagate essentially parallel to each other to a friend. Using these measures, the “wave” of the currently decoded parts (such as p1, p4, p5 + 1 and pk in this figure) of the entropy slices can generally be located along a line forming an angle to the direction of the entropy slice, which is smaller than the direction serving sequences. The wave will probably have to have a 1y to 2x slope for the block positions so that the upper left support element is always there for all threads processing slices in parallel.

The decoder can be configured to perform entropy decoding for each entropy slice along the corresponding entropy coding path in units of portions of the sample array of the corresponding entropy slice so that the entropy slices consist of the same number of parts, respectively, and the sequence of portions along the paths of the entropy slice was aligned relative to each other in a direction transverse to the trajectories of the entropy slice. The current portion of the portion of the predefined entropy slice belongs to the resulting grid of parts (for example, LCUs or macroblocks). When entropy decoding is performed for each entropy slice along the corresponding entropy coding path, the decoder can save the offsets / shifts between the initial times between the directly following successive entropy slices in the order of the entropy slices so that the wave of the currently decoded parts of the entropy slices forms a diagonal such as the corresponding line with a slope of the positions of 0.5x blocks with respect to the trajectory of the entropy slice and the direction in order ntropiynyh slices. The offset / shift can correspond to two parts for all pairs of immediately following entropy slices. Alternatively, the decoder may simply prevent the entropy slices from becoming smaller than two parts between the currently decoded portions immediately following one after another (and directly bordering while the portions of the 12 array of samples are considering them). See the figure above: as soon as the part / block p4 is decoded, the part / block to the right of it is decoded according to the order of 16 paths, and at the same time, if there is, X is decoded or any of the previous parts / blocks). Using these measures, the decoder can use probability estimates that are already adapted based on the contents of part / block p4, that is, the part within a spatially adjacent portion aligned with the portion of portion 12 of the predetermined entropy slice following the current part X in order of 16 paths for determining probability estimates that will be used when decoding X. In the case of constant bias when decoding the directly following entropy slices of two parts, the decoder can the ability to use probability estimates that are already adapted based on the contents of the p4 part / block at the same time to entropy decode the next part (i.e. the part to the right of p4) of the spatially adjacent entropy slice.

As described above, weighted summation of already adapted probability estimates can be used to determine the probability estimates that will be used to decode X.

As also described above, the order of the entropy slices can also pass through the boundaries of the frame.

It should be noted that the just stated borrowing of probability from previous entropy slices can be performed for each part of the current / predetermined entropy slice for which such neighboring parts are available in previous entropy slices. That is, this is also true for the first part along the direction 16 of the trajectory, and for this very first part / block (leftmost in each entropy slice in this figure), the borrowing equals the initialization described above.

For better adaptation, also in this case, 2 of the above methods can also be combined together. The results of this process with 1 and 4 streams, i.e. in parallel using processing units, are illustrated in FIG. 10-13 (2LCU + Prob.Adap or 2LCU graphs).

For a better understanding of the context of the aforementioned embodiments, and in particular the additional embodiments described below, namely the use of LCUs, you can first look at the H.264 / AVC structure.

An encoded video sequence in H.264 / AVC consists of a sequence of access units that are collected in a stream of NAL units and they use only one set of sequence parameters. Each video sequence can be decoded independently. An encoded sequence consists of a sequence of encoded images. An encoded frame may be an entire frame or a single field. Each image is divided into fixed-sized macroblocks (in HEVC: LCU). Several macroblocks or LCUs can be combined together into one slice. An image is therefore a collection of one or more slices. The purpose of this data separation is to enable independent decoding of samples in the image area that is represented by this slice, without using data from other slices.

The technique, also often referred to as “entropy slices,” is the division of a traditional slice into additional sub-slices. In particular, this means splitting the entropy encoded data of a single slice. The location of the entropy slices in the slice can have various varieties. The simplest variation is to use each line of LCU / macroblocks in a frame as one entropy slice. Alternatively, columns or individual regions can be used as entropy slices, which can even be interrupted and switched with each other, for example, as slice 1 in FIG. fourteen.

The obvious goal of the entropy slice concept is to enable parallel use of the CPU / GPU and multi-core architecture to improve the decoding process time, that is, to speed up the process. The current slice can be divided into sections that can be analyzed and restored independently of other data in the slice. Although some advantages can be achieved using the entropy slice approach, there are some disadvantages.

First of all, the main goal is to create a bitstream that is suitable for the process of parallel encoding and decoding. Keep in mind that LCUs can only be encoded if neighboring LCUs (left, top, top-right) are already available as an encoded version to use spatial and motion information for prediction. To allow parallelism using separation, a shift between slice processing (e.g., a shift of 2 LCUs, which is typical of a wave approach) should be implemented. Due to the adaptation of the CABAC probabilities, the LCU uses the probabilities that are available from the previous coded LCU. Regarding the order of the raster scan, the problem that arises with the separation of the image is to avoid parallelism, since the first LCU of each line depends on the last LCU of the previous line. This is influenced by the fact that the dependencies of CABAC probabilities between slices must be removed so that several slices can start at the same time. One way to do this is to simply re-initialize CABAC, whereby, however, all borrowed data will be lost. As a result, the bit rate may increase.

Each slice then generates its own bitstream substream, which can be placed sequentially in the main stream. But it is necessary to transmit special information to the decoder so that these slices and their positions in the main stream can be correctly identified. Two signaling scenarios are supported. Location information can be stored in the image header (slice length information) or in each slice header (points, as the beginning of the code). Byte alignment at the end of each entropy slice and location information increases the loss.

To eliminate the shortcomings introduced by signaling for entropy slices, it is important to use a good coding technique for signaling. A significant disadvantage of signaling the entropy slice is introduced into the frame if the initial codes are used for each slice, that is, too many additional bytes (for example, at least 4 bytes per slice) are added to the bitstream. Of course, inserting entropy slices 90 using the initial codes has an advantage in low latency scenarios where the encoder should be able to instantly output entropy slices. In such cases, there is no way to pre-transmit entry point information.

In less stringent low-latency scenarios, however, the ability to preserve the slice length (offsets) seems more appropriate. One known method for encoding such information is variable-length coding (VLC) or Golomb exponential coding. The main feature of VLC is the addition of empty information (zeros) before real information. Using these zeros, a code can be defined that stores information about the offset length. The authors propose another technique for implementing this, a diagram of which is shown in FIG. 15, where = EntropySliceSize is the number of bytes contained in the entropy slice. Each subsequent X (offset) is defined as the difference in size between the previous already encoded and transmitted offset of the entropy slice and the current slice. The main features of this concept are formed from difference values, depending on the size of X, so the amount of data can be reduced to 3 bits, which makes it possible to extract accurate information about the size of each entropy slice using a decoder. Compared to VLC, a reduction in the number of bits in the header of the entropy slice can be obtained.

That is, in accordance with an aspect of FIG. 15, for example, provides the concept of entropy coding of an information signal, which allows a higher degree of compression, despite the possibility of parallel processing, compared with currently available concepts. In accordance with this aspect, the entropy encoded section 20 of the data stream into which the sample array is encoded contains entropy slices, such as 90 in FIG. 8, in which various portions 12 of the sample array are entropy encoded, and the header illustrated by dashed lines 300 in FIG. 9 contains information that shows the initial positions 302 of the entropy slice 90, measured in the region of entropy decoding, within the entropy encoded data stream, this information contains a difference value for the predetermined entropy slice, revealing the difference between the initial position of the previous entropy slice n-1 and the initial By the position of the predetermined entropy slice n, the data stream includes data difference values, as a sequence of VLC bits.

In particular, the VLC bit sequence may have a variable length prefix y, indicating that the difference x falls into the y-th sequence of the number of z intervals [0.2 ^a -1], [2 ^a , 2 ^b +2 ^a -1], [ 2 ^b +2 ^a, 2 ^b +2 ^c +2 ^a -1], and so on, and the encoded via PCM suffix y-th sequence of lengths a, b, c ... If a, b, c ... are selected so that they are a power of two, if the corresponding y is added, that is, so that a + 1, b + 2, c + 3 and so on, all are powers of two, then byte alignment can be preserved. The number z is not limited to three, as selected by way of example in FIG. fifteen.

An encoder, such as the encoder of FIG. 1, respectively, is designed to convert the difference of consecutive starting positions into a sequence of VLC bits, that is, by first determining the prefix, that is, in which the difference (y-th) is in the sub-interval, and then setting the suffix plus the difference, the starting position minus the y-th of 0 , 2 ^a , 2 ^b +2 ^a and so on. A decoder, such as in FIG. 6, respectively, is designed to obtain the initial position of the current entropy slice n from the VLC bit sequence, that is, by first examining the prefix to obtain y, then setting the difference value for the suffix value plus the yth 0, 2 ^a , 2 ^b +2 ^a and so on, and then adding the difference for the starting point of the previous entropy slice n-1.

By segmenting the entropy slices, an additional advantage can be provided, namely, low-latency transmission and decoding acceleration.

In the video stream, providing the possibility of higher resolutions (Full-HD, QUAD-HD, etc.) leads to more data to be transmitted. For time-sensitive scenarios, the so-called low-latency use cases (<145 milliseconds), transmission time becomes a critical factor. Consider an ADSL data line for video conferencing applications. In this case, the so-called random access stream points, usually referred to as I-frames, will be candidates for creating a bottleneck during transmission.

To solve this problem and minimize transmission delay and decoding time, i.e. end-to-end delay, a new technique can be applied to the entropy slices interleaving scheme for parallel transmission and processing.

HEVC provides the so-called wave data processing on the side of the decoder. They are going to provide it with the help of entropy slices. In the usual case, the data of the entire slice is delivered at one time. The decoding process begins as soon as the encoded data arrives at the wave decoder facility. In order to reduce the time that a decoder can start and end a frame, the segmentation of entropy slices into small fragments using the interleaving approach is used in accordance with the present embodiment. Therefore, the encoder can deliver data corresponding to a specific entropy slice to the transport layer earlier than in the normal case. In this case, this leads to faster transmission and earlier start of the parallel decoding process in the client.

Creating slices fragments can be additionally ensured by dividing the entropy slice into additional slices, while preserving all the dependencies (dependent slices). If this is done in each largest coding unit (LCU) / coding tree unit (CTU), then these fragments can be interleaved, additionally using system-level methods that take into account the transmission of fragments in interleaved form and restoration, or at least providing information about the original decoding order fragments through additional signaling. Such signaling may be a decoding sequence number (DON), such as the IETF RTP payload format for H.264 / AVC (RFC 3984). Another systemic method may consist in applying fragments of the wave substream to another transport stream, as in MEPG-2 systems, by assigning different PIDs for each of them, then multiplexing and thereby interleaving them in the transport channel.

This approach can also be applied across the boundaries of the frame if the slice (s) of the next frame or the entropy slice (s) can already be decoded, for example, by a wave method, based on the knowledge of the requested information for decoding the entropy slice of the next frame due to the availability of interframe links. This data, which can already be decoded, of a frame following the decoding order, can be obtained from the maximum allowed / transmitted length of the motion vector or additional information in the stream indicating the dependence of the data parts on the previous frame (s)), or from a fixed link scheme, indicating a position transmitted in a sequence-fixed position, such as a set of parameters. This will be further described below.

An image may be encoded with one entropy slice per line (s) of largest coding units (LCUs), as shown in FIG. 2 and 9. It is preferable to use wave methods on the side of the decoder. During the encoding process, the bitstream of each slice can be divided into segments of constant size. Then the resulting segments are interleaved and can be transmitted for transmission. The constant size of the segments can cause a problem at the end of the bitstream due to its variable length.

There are two possible generalized solutions. The first solution is to generate single-byte segments (usually a bitstream representing a slice is aligned with a byte), and control the consumption of bytes using each decoder tool, that is, the decoder determines the end of the entropy slice.

The second is to use the completion code at the end of the slice. In the latter case, variable-length segments may be used, but this may also lead to more data. Another way is to transfer the length of the entropy slice. One such alternative is described below.

The segment size and interleaving method can be transmitted either in a single SEI message or in an SPS.

The transmission scheme is shown in FIG. 16.

Thus, in accordance with an aspect of FIG. 16, for example, a concept of entropy coding of an array of samples is provided, which provides lower latency compared to currently available concepts. In accordance with this aspect, the encoded data stream 20 into which the information signal is encoded contains slices, such as entropy slices 90 or simply completely independently decoded slices (shown on the left), in which various portions 12 of the information signal (with prediction and / or entropy) are encoded, and these slices 90 are divided into fragments (shaded rectangles 310), which are located in the encoded data stream 20 in an intermittent manner (shown on the right side), interleaving indicated by bracket 312.

As indicated above and described with respect to other aspects, the slices may be entropy slices 90, which, in turn, may be corresponding subsets of frame slices, and accordingly, the encoded data stream may be entropy encoded data stream 20.

The interleaving 312 of slices 90 allows a lower delay, since the decoder responsible for decoding any of the slices 90 does not have to wait for the length of time spent on previous slices of the other decoders (according to the order of 16 slices). Instead, all available decoders have the ability to start decoding the slices associated with them as soon as the first fragment 310 is available, and the potential dependencies between the parts are determined, compared with the wave approach.

Different portions 12 can be entropy encoded into entropy slices 90, using probability estimates that are set independently among entropy slices, or using borrowing between entropy slices of probability estimates, as described above, for example, in which different portions 12 are encoded into entropy slices 90 along the corresponding trajectories 14 of entropy coding, and dependent entropy slices have corresponding portions 12 encoded in them using probability estimates obtained including after the probability estimates used in entropy slices of a higher rank in spatially adjacent parts of the corresponding portion, as described previously.

The entropy encoded data stream 20 may further comprise a header, as shown as an additional feature in FIG. 16. The header 300 may be associated with a frame (array of samples) of the sequence 30, the header 300 contains information that shows the length of the entropy slices. Information regarding the length of the entropy slices 90 may be encoded within the header 300, as described above, using VLC codes. Using knowledge of the length of the entropy slices, on the decoding side, the last fragment associated with each of the entropy slices 90 and its length can be identified. However, start codes or other indication schemes may also be used. The initial positions of the slice can also be identified simply by a decoding process that knows the completion of the slice. Therefore, it may be possible to simply rely on instructions from the decoder, but this requires signaling between the decoders and in some cases, if the “earlier” entropy slice ends later than the “later” slice in the stream. This may require, for certain cases, “adaptive” signaling in the stream, which may be based on initial codes.

For example, entropy slices 90 may be located after header 300, as illustrated in FIG. 16.

Fragments 310 may be of equal length, at least as far as the initial adjacent portion of the sequence of fragments 310 from the first of the fragments in the order in which the fragments are located in the entropy encoded data stream 20. Subsequent fragments may vary in length. Subsequent fragments may be equal to or less than the length of the initial adjacent portion of the sequence. The length of the subsequent fragments can be obtained from the above information in the header 300, which reveals the length of the entropy slices 90 or the position of the beginning. Fragments 310 may be arranged in the entropy-encoded data stream 20 cyclically in accordance with the order defined among the entropy slices. For entropy slices, fragments of which are completely in previous cycles, they can be skipped in the current and subsequent cycles.

Other information signals than the sequence of the array of samples, such as a video signal, can also be transmitted through the data stream 20. The various portions 12 thus need not be portions of a predetermined array of samples, such as an image / frame.

As described above, entropy slices 90 may have different portions 12 of an array of 10 samples encoded therein by prediction coding, using prediction between entropy slices and / or inter-frame prediction, entropy encoding of prediction errors to predict between entropy slices and / or inter-frame prediction. That is, as described above, different portions can be spatially different portions of frame 10 or multiple frames 30. The latter case applies if the slice (s) of the next frame (s) or the entropy slice (s) can already be decoded, for example, by the wave method based on the knowledge of the requested information for decoding the entropy slice of the next frame due to the available inter-frame links. Already decoded data of a frame following the decoding order can be obtained from the maximum allowed / transmitted length of the motion vector or from additional information in the stream indicating the dependences of the data parts on the previous frame (s)), and the prediction between the entropy slices may involve internal prediction, while inter prediction may involve motion compensation prediction. An example is outlined below.

The aforementioned independence of the probability estimate setting among entropy slices can apply to probability adaptation as well as content modeling. That is, the content selected within the entropy slice can be selected independently from other entropy slices, and the probability assessment of the content can also be initialized and adapted independently of any other entropy slice.

The corresponding entropy decoder can be constructed as follows.

An entropy decoder configured to decode an entropy encoded data stream 20 that contains entropy slices 90 into which various portions of a frame are entropy encoded, the entropy slices being subdivided into fragments 310 that are arranged in an entropy encoded data stream 20 in an intermittent manner, as configured shown in FIG. 6, and may further comprise a de-interleaver configured to de-interleave fragments 310, shown by symbol 314 in FIG. 16.

In particular, as illustrated in FIG. 5, an entropy decoder may comprise a plurality of entropy decoders 130, such as threads running in different processing cores, with the deinterleaver being configured for each entropy slice to direct fragments 310 thereof to the entropy decoder 44 associated with the corresponding entropy slice.

In other words, the entropy slices can be divided into fragments, which, in turn, can be interleaved, and the decoder may contain a deinterleaver to deinterleave the fragments and can begin to process entropy slices in parallel along trajectories 16 even before receiving any of the entropy slices as a whole. It should be remembered that the length of the fragments is preferably measured in the region of entropy coding, and not in the syntactic region so as to correspond, for example, to the set of certain spatial parts / blocks in the image and the like, although the latter option will also be available.

The following describes the possible use of time dependencies. The same can be used in addition, or alternatively to the already described options for improving the probability assessment.

The wave data processing sample, as described in this case, can be extended to entropy coding with a new calculation of probabilities for each LCU, so as to use the time dependencies between frames.

As you know, probabilities will be reinitialized at the beginning of each frame (first LCU). Thus, the probabilities that were already obtained in the previous frame are lost. To reduce the loss of coding efficiency, it is possible to transmit the final state of the image (compare 320) or, in the case of using the entropy slice, the final state of the slice (compare 322) of probabilities from the reference frame 324 to the first LCU 50 of the current frame 10 or entropy slice 12, respectively (Fig. . 17). Such data of the corresponding slice in the reference frame can be obtained not only in the final position, but also in the previous position in the reference slices, since parallel wave processing of the data can also pass through the frame boundaries, that is, while the frame slice is encoded, the encoding process of the previous slice frame still may not be finished. Therefore, an alarm can be used to indicate a reference position, or it can be indicated by a diagram.

Using the above remark, it is thus possible to initialize in the initial phase 84 and 234, respectively, to set P _{{i, j}} , and {i, j} denotes the first CU 50 in the kth entropy slice of the current array of samples equal to, or at least depending on any T (P _{{i, j} '} ), and {i, j}' denotes CU within the previous (in the order of coding of the array of samples, which may be equal to the order of presentation) array of samples, or a combination of several T (P _{{i, j} '} ). This can be done only for k = 0, or for each entropy slice k∈ {1 ... K}, and K denotes the number of entropy slices in the current frame. Temporary initialization can be done additionally or alternatively to the spatial initialization described above. That is, P {i, j}, where {i, j} denotes the first CU 50 in the kth entropy slice, can be set equal to some combination (such as the average value of sme) T (P _{{i, j} '} ) and T (P _{{i, j} spatial} ), where {i, j} 'denotes the CU within the previous (previously decoded) array of samples or a combination of several T (P _{{i, j}'} ), and {i, j} _spatial denotes CU within the previous entropy slice of the current array of samples. While {i, j} 'is being considered, P _{{i, j}} , and {i, j} denotes the first (in order of 14 entropy coding) CU 50 in the kth (in order of 14 entropy coding) entropy slice of the current array of samples, can be set equal to T (P _{{i, j} '} ), and {i, j}' denotes the last (in order of 14 entropy coding) CU within the kth (in order of entropy slices) entropy slice in the previous (in order encoding an array of samples) an array of samples or the last CU within the last (in order of entropy slices) entropy slice in the previous cm (in order of coding the samples arrays) array samples. Again, this temporary initialization can only be performed for the first entropy slice in the sample array.

The process of analyzing the final state of the reference frame was tested using the method of adapting probabilities, these results are illustrated in FIG. 10 - FIG. 19 (timeline).

Another possibility of using data from other frames is to exchange the obtained probabilities between the associated LCUs. The main idea is based on the assertion that the properties of the reference frame are not very different from the current frame. To speed up the study of probabilities by LCU in a frame, you can try to transfer the final state of each LCU to the corresponding LCU in the current frame. This proposal is illustrated in FIG. eighteen.

A reference frame can be understood as various possibilities. For example, the frame that is last encoded can be used as a reference frame. Otherwise, the last encoded frame only from the same time level can be assigned as a reference.

In addition, this approach can be combined with the methods already proposed above, such as using the last (slice) of information from the reference frame, adapting probabilities, and using the second LCU of the upper line.

The above spatial adaptation process may be modified so that it is

,

,

where N can be 1 or more than 1 and {i, j} _{1 ... N is} selected from (is within) any previous (in order of 16 entropy slices) entropy slice 90 in the current array of 10 samples and the associated portion 12, accordingly, M can be 1 or more than 1 and {i, j} ' _{1 ... M} is within the previous array of samples 350. It may happen that at least one of the CU 50 {i, j}' _{1. ..M is} combined with p (n). Regarding the possible choices CU 50 {i, j} _{1 ... N} reference is made to the above description. The “mean value” function may be one of a weighted summation, a median function, etc.

The above spatial adaptation process may be replaced by

,

,

where M can be 1 or more than 1 and {i, j} ' _{1 ... M} is within the previous array of samples. It may happen that at least {i, j} ' _{1 ... M is} aligned with p (n). Regarding the possible choices {i, j} _{1 ... N} reference is made to the above description. The “mean value” function may be one of a weighted summation, a median function, etc. It may happen that at least {i, j} ' _{1 ... M is} aligned with p (n).

As a specific extension to the use of related information, an approach may be used to use received data from other blocks from one or even more reference frames.

The previously mentioned methods use the information received only from immediate neighbors in the current frame or in reference frames. However, this does not imply that the probabilities obtained are in this case the best. Adjacent LCUs, according to image partitions (difference information), do not always have better probability models. It is contemplated that better results can be achieved using the blocks on the basis of which the prediction will be performed. And thus, this corresponding unit can be used as a reference for the current LCU.

Thus, in the above adaptation example, {i, j} _{1 ... N} and / or {i, j} ' _{1 ... M} can be selected depending on the use of CU as sources of prediction parameters for p (n).

The presented adaptation / initialization schemes of time probability can also be used without entropy slices or with one entropy slice in a frame.

In accordance with the latter aspect, an increase in the probability adaptation rate is achieved by combining probability adaptation in the time domain of adjacent / related frames. Described is a decoder, such as the decoder in FIG. 6, and this decoder is configured to restore a sequence of arrays of samples from a data stream of an entropy encoder, and is configured to entropy decode the current frame of a data stream of an entropy encoder to restore the current array of samples of a sequence of arrays of samples, perform entropy decoding along the path of entropy encoding and use probability estimates and adaptation probability estimates along the trajectory of entropy coding, using previously decoded a portion of the current frame, wherein the entropy decoding stage is configured to initialize or determine a probability estimate for the current frame based on probability estimates used in decoding a previously decoded frame of the entropy encoded data stream.

That is, for example, the probability estimates for the current frame are initialized based on the probability estimates resulting from the completion of decoding of a previously decoded frame of an entropy encoded data stream. Therefore, buffer requirements are low, since it is just that the final state of the probability estimates must be buffered before decoding of the current frame begins. Of course, this aspect is combined with the aspect of FIG. 1-9 in that for the first parts of each portion 12, not only probability estimates used for spatially adjacent parts in previous entropy slices (if available) are used, but also in a weighted way, for example, the final state of probability estimates (for example, spatially) corresponding entropy slice in the previous frame. Such data of the corresponding part in the reference frame can be obtained not only in the final position, but also from the previous position in the reference slices, since parallel wave processing of the data can also pass through the frame boundaries, that is, when the frame slice is encoded, the slice encoding process of the previous the frame may not be finished yet. Therefore, an alarm can be used to indicate a reference position, or it can be indicated by a circuit.

Additionally, for example, the probability estimates used to encode parts / blocks of a previously decoded frame are buffered everything, not just the final state, and the decoder will entropy decode a predetermined entropy slice (with reference to the above description of spatially combined probability acquisition), perform entropy decoding of the current part (X) of the predefined entropy slice, based on the corresponding probability estimates of the predefined entropy slice, which adapt are used using the previously decoded part of the predetermined entropy slice (including p1, for example) and on the probability estimates that are used in the entropy decoding of the spatially corresponding part of the entropy slice in the previously decoded frame, if desired, additionally using the probability estimates that are used in the entropy decoding spatially adjacent, in the order of entropy slices of the previous entropy slice (a slice containing X, for example) in the neighboring part (such both p4) spatially neighboring entropy slice, as described above. As also described above, the spatial correspondence between the parts and the identification of the corresponding part for borrowing the probability for the current frame among the previously decoded frame can be determined using motion information such as motion indices, motion vectors, etc., of the current part / block.

Until now, the wave propagating during wave processing of the data has been primarily described as being propagated obliquely through one array of 10 samples, and encoding / decoding is performed one array of samples after another. However, this is not a necessity. Reference is made to FIG. 19. FIG. 19 shows a portion of a sequence of sample arrays, wherein sequence arrays of samples are defined among it and depicted as arranged in order 380 of encoding of sample arrays, which may or may not coincide with the presentation time order. FIG. 19, for example, shows the division of arrays of 10 samples into four entropy slices each. Already encoded / decoded entropy slices are shown hatched. Four coding / decoding streams (coding / decoding cascade) 382 are currently performed on these four entropy slices 12 of the sample array having index n. However, FIG. 19 shows the remaining stream number 5, and it is possible that this additional stream 382 having number 5 in FIG. 19, works to encode / decode the next array of samples on the line, i.e., n + 1, in portions for which it is guaranteed that the corresponding reference portions in the currently encoded / decoded frame n are already available, that is, already processed by any of the streams 1 -four. These portions are reference, for example, in the predictions shown as 64 in FIG. one.

FIG. 19 shows by way of example a dashed 384 line extending through an array of n + 1 samples that is already aligned with the boundary between the processed, i.e., encoded / decoded portion of the array of n samples, i.e., the shaded portion within the n array of samples on one side, and not yet processed portion, that is, an unshaded portion of an array of n samples on the other hand. Using the bidirectional arrows of FIG. 19 also shows the maximum possible length of motion vectors measured in the column and row directions, i.e., y _max and x _max , respectively. Accordingly, FIG. 19 also shows, with a dash-dotted line 386, a displaced version of line 384, namely, line 386, which is spaced apart from line 384 at the minimum possible distance so that the distance does not fall below y _max in the direction of the column and x _max in the direction of the line. As can be seen, there are 50 coding units in an array of n + 1 samples for which any supporting portion in an array of n samples is guaranteed to be found as being completely contained within the already processed portion of this array of n samples, namely those that are located in half of the array of n + 1 samples located on the upper side relative to line 386. Accordingly, stream 5 already has the ability to work to decode / encode these coding units, as shown in FIG. 19. As can be seen, even the sixth stream can process the second entropy slice in order of 16 entropy slices of an array of n + 1 samples. In this regard, the wave propagates not only spatially, but also in the time domain through the space-time space filled with a sequence of 30 arrays of samples.

It should be noted that the just-mentioned wave aspect also really works in combination with the above relationships of probability estimates along the boundaries of the entropy slice.

Additionally, regarding the foregoing aspect of the fragment, it should also be noted that the separation of entropy slices into smaller parts, that is, fragments, is not limited to performing in the field of entropy coding, that is, in the entropy compressed region. If you think about the discussion above: the entropy slices described above have the advantage of reducing the loss of coding efficiency despite the possibility of wave data processing due to the origin of probability estimates from previously encoded / decoded entropy slices of the same or previously encoded / decoded frame, i.e. initializing and / or adapting probability estimates based on probability estimates of such previous entropy slices. Each of these entropy slices is assumed to be entropy encoded / decoded by a single stream in the case of wave data processing. That is, when the entropy slices are shared, it is not necessary to process fragments that can be encoded / decoded in parallel. Instead, the encoder is simply provided with the ability to output sections of the bitstream of its entropy slice before the completion of entropy encoding, and the decoder is provided with the ability to process these sections, i.e. fragments, before receiving the remaining fragments of the same entropy slice. In addition, interleaving should be allowed on the receiving side. To allow the last de-interleaving, however, separation in the field of entropy coding is not required. In particular, it is possible to perform the above separation of entropy slices into smaller fragments without serious loss of coding efficiency, periodically simply reloading the internal state of the probability interval, i.e., the value of the width of the probability interval, and the offset value, respectively, of the entropy encoding / decoding core. Probability estimates, however, are not reloaded. Instead, they are continuously updated / adapted from the beginning to the end of the entropy slices, respectively. Using these measures, it is possible to separate entropy slices into separate fragments, and the separation is performed in the field of syntactic elements, and not in the area of a compressed bit stream. Separation may be the result of spatial separation, as described below, to facilitate the transfer of boundaries between fragments to the decoder. Each of the fragments can be provided with its own fragment header, which reveals, for example, its initial position in the array of samples, measured, for example, in coding order 14 relative to the initial position of the corresponding entropy slice along with the index of its entropy slice, or relative to the known location of the array of 10 samples such as the upper left corner.

For a clearer description of the division of entropy slices into fragments in accordance with the last embodiment, reference is made to FIG. 20. FIG. 20 shows, simply for illustrative purposes, an array of 10 samples, as subdivided into four entropy sequences. The currently encoded portions of the array of 10 samples are shown as hatched. Three streams currently perform entropy encoding of an array of 10 samples and output fragments of entropy slices based on direct consideration: see, for example, the first entropy slice in order of 16 entropy slices, which corresponds to a portion 12 of an array of 10 samples. After section 12a of portion 12 is encoded, the encoder generates a fragment 390 from it, that is, the entropy encoding core 80 performs some final procedure to complete the arithmetic bit stream obtained from subport 12a in the case of arithmetic encoding to form fragment 390. The encoding procedure is then resumed with respect to the subsequent subportion 12b of the entropy slice 12 in coding order 14, starting a new entropy bitstream. This implies, for example, that internal states, such as the value of the width of the probability interval and the value of the offset of the probability interval of the entropy coding kernel 80, are reloaded. Probability estimates, however, are not reloaded. They are left unchanged. This is illustrated in FIG. 20 using arrow 392. FIG. 20 as an example, it is shown that the entropy slice or portion 12 is divided into more than two subportions, and accordingly, even the second 1b fragment must be somewhat entropically completed before reaching the final portion 12 in encoding order 14, after which the next fragment on the line begins, and so on. d.

At the same time, another thread processes the second entropy slice or portion 12 in order of 16 entropy slices. After the first subportion of this second entropy slice / portion 12 has ended, fragment 2a is output, after which entropy encoding of the remaining part of the second entropy slice begins, while preserving, however, the probability estimate that is valid at the end of fragment 2a.

Using the time axis 394 of FIG. 20 illustrates that fragments 390 are output as soon as they are completed. This results in an interleaving similar to the interleaving shown in FIG. 16. Each fragment can be packetized into a packet and transported to the decoding side through a transport layer in any order. The transport layer is illustrated using arrow 396.

The decoder must reassign fragments to subports 12a, 12b, etc. For this purpose, each fragment 390 may have a header section 398, which reveals the initial location of the associated subport 12a or 12b, that is, a subport of syntax elements describing the same thing is entropy encoded in the corresponding fragment. Using this information, the decoder has the ability to associate each fragment 390 with its entropy sequence and with its subportion within portion 12 of this entropy slice.

For the purpose of illustrating FIG. 20 also shows by way of example the possibility when it is not required that the transition between successive subportions 12a and 12b of the entropy slice 12 coincide with the boundary between successive coding units 50. Instead, this transition may be defined at a deeper level of the above-mentioned exemplary division of coding units into several trees. The location information contained in headers 398 may indicate the start of the subportation associated with the current fragment 390, accurately enough to identify the corresponding sub-block of the corresponding coding block, that is, the location within the sequence of syntax elements that describes the corresponding sub-block.

As it became clear from the discussion above, the almost no loss of coding efficiency is the result of dividing the entropy slices into fragments. Simply entropy termination and packaging processes may involve some loss in coding efficiency, but on the other hand, reducing latency is enormous.

Again, it should be noted that the spatial aspect of fragmentation just mentioned also really works in combination with the above compounds of probability estimation along the boundaries of the entropy slice, in the spatial and temporal domain.

A decoder, such as the decoder in FIG. 6 may cancel fragment transfer as follows. In particular, the decoder can check to which entropy slice the current fragment belongs. This check may be performed based on the above location information. Then, it can be checked whether or not the current fragment of the first portion of the portion of the corresponding entropy slice corresponds along the entropy coding path 14. If yes, then the decoder can entropy decode the current fragment when adapting the corresponding probability estimates and take into account the state of the corresponding probability estimates that are found at the end of the entropy decoding of the current fragment when another fragment is entropically decoded, which corresponds to the second subportion of a portion of a predetermined entropy slice along the path of the entropy coding. “Accounting” may involve settling probability estimates at the beginning of fragment 1b, equal to probability estimates, detecting them directly, by adapting the probability, starting from the state of the probability estimate at the beginning of fragment 1a, at the end 1a of the subportation of fragment 12a, or equal to combining it with entropy estimates probabilities from other entropy slices, as described above. While the initialization of probability at the beginning of the first fragment 12a is considered, reference is made to the above discussion, since it also forms the beginning of the corresponding entropy slice. In other words, if the current slice is the second or subsequent fragment in order of 14, then the decoder can entropy decode the current fragment using probability estimates that depend on the probability estimates found at the end of the entropy-decoded fragment, which corresponds to a subportation of a portion of a predetermined entropy slice preceding a subportation corresponding to the current fragment along the path 14 of entropy coding.

The above description shows various methods that can be used for parallel coding and decoding, as well as used to optimize existing processes in the promising HEVC video coding standard. A short overview of entropy slices was presented. It is shown how they can be formed, what advantages can be provided by slicing, and what disadvantages can result from these methods. Many methods have been proposed that are supposed to improve the process of studying probabilities by LCU (largest coding unit) in a frame by making better use of local dependencies between LCUs, as well as time dependencies between LCUs of different frames. Various combinations are argued to provide improvements for both concepts with and without parallelization of coding and decoding.

The increase in efficiency with high-performance coding, for example, using the best combination of the proposed approaches, is -0.4% for internal, -0.78% for low latency and -0.63% for random access compared to HM3.0 without using entropy slices or -0.7% for internal, -1.95% for low latency and -1.5% for random access compared to the entropy slice approach with conventional re-initialization.

In particular, the following methods were presented above, including:

for using not only local, but also time dependencies of LCUs to optimize the adaptation of CABAC probabilities before encoding each LCU, see FIG. 1-9, 17 and 18;

to provide greater flexibility in decoding, also entropy slices can be used so that certain areas in the frame become independent of each other;

to enable minimal signaling of the initial position of the slice / entropy slices for parallelization, for example, wave data processing, see FIG. fifteen;

to enable low-delay transport in a parallelized environment, the encoder-transmitter-receiver-decoder by interleaving the entropy slices / slices, see FIG. 16.

All of the methods mentioned above were combined and tested in HM3.0. The obtained results, where the control point is HM3.0 without any entropy slice embodiment, are presented in Tables 1 and 2 (where 2LCU is the use of the second LCU of the upper line; 2LCU + Prob.Adap is a 2LCU method combined with a probability adaptation method; related to time - use of time dependencies (final state of the reference frame) with adaptation of probabilities for each LCU).

Table 1 Generalized RD results with 1 thread 1 thread 2LCU 2LCU + Prob.Adap Prob.Adap Time related Interior 0.14 -0.31 -0.4 Low latency 0.66 0.09 -0.12 -0.78 Random access 0.48 -0.1 -0.24 -0.63

table 2 4-thread generalized RD results 4 threads 2LCU 2LCU + Prob.Adap Prob.Adap Time related Interior 0.19 -0.27 -0.27 Low latency 1.01 0.54 0.63 0.17 Random access 0.74 0.2 0.24 0.01

However, it is interesting to know how the proposed approaches involve wave data processing with reinitialization of probabilities at the beginning of each LCU line. These results are illustrated in Tables 3 and 4 (where orig_neiInit is a comparison of HM3.0 without using entropy slices using entropy slices with reinitialization).

Table 3 Generalized RD results with 1 thread. The reference element is the new initialization. 1 thread orig_
neiInit 2LCU 2LCU + ProbAdap Probadap Time related Interior 0.3 -0.15 -0.62 -0.7 Random access 0.9 -0.4 -0.99 -1.13 -1.51 Low latency 1.19 -0.51 -1.08 -1.29 -1.95

Table 4 Generalized RD results with 4 threads. The reference element is the new initialization. 4 threads orig_
neiInit 2LCU 2LCU + ProbAdap Probadap Time related Interior 0.34 -0.14 -0.61 -0.61 Random access 1.18 -0.43 -0.96 -0.96 -1.15 Low latency 1,61 -0.5 -0.92 -1.05 -1.41

The above results show that a significantly greater use of dependencies within and between frames and the rational use of information already received prevents averaged losses.

The HEVC video wave data processing approach for encoding and decoding combines the ability to use dependencies between adjacent LCUs as well as the temporal dependencies of a frame with the concept of parallel wave processing. In this way, losses can be reduced, and an increase in efficiency can be provided.

An increase in the probability adaptation rate is provided by calculating the adaptation of the probability of spatially adjacent entropy slices.

As already mentioned above, all of the above aspects can be combined with each other, and thus the mention of certain features of the embodiment in relation to a certain aspect should, of course, also apply to other aspects.

Although some aspects are described in the context of the device, it is clear that these aspects also represent a description of the corresponding method, where the unit or device corresponds to a method step or a feature of a method step. Similarly, the aspects described in the context of a method step also provide a description of the corresponding unit or element or feature of the corresponding device. Some or all of the steps of the method can be performed (or used) using a hardware device, such as a microprocessor, programmable computer, or electronic circuit. In some embodiments, one or more of the most important steps of the method can be performed using such a device.

The above-mentioned inventive encoded signals may be stored on a digital storage medium or may be transmitted over a transmission medium, such as a wireless transmission medium or a wired transmission medium, such as the Internet.

Depending on the specific requirements of the embodiment, embodiments of the invention may be embodied in hardware or in software. Embodiment can be performed using a digital storage medium, for example, a floppy disk, DVD (digital versatile disk), Blu-ray, CD (compact disk), ROM (read-only memory), ROM (programmable ROM), EPROM (erasable programmable ROM) , An EEPROM (electrically erasable programmable ROM) or flash memory having electronically readable control signals that interact (or can interact) with the programmable computer system in such a way that the corresponding s the way. Therefore, the digital storage medium may be a computer readable medium.

Some embodiments of the invention comprise a storage medium having electronically readable control signals that can interact with a programmable computer system in such a way that one of the described methods is performed.

In general, embodiments of the present invention may be embodied as a computer program product with program code, the program code being used to perform one of the methods when the computer program product is executed on a computer. The program code may, for example, be stored on a computer readable medium.

Other embodiments include a computer program for performing one of the methods described, stored on a computer-readable medium.

In other words, an embodiment of the invented method is therefore a computer program having program code for executing one of the methods described when this computer program is executed on a computer.

An additional embodiment of the invented methods is therefore a storage medium (either a digital storage medium or a computer readable medium) comprising a computer program recorded thereon for performing one of the described methods. A storage medium, a digital storage medium or a recorded medium is usually tangible and / or non-temporary.

An additional embodiment of the invented method is therefore a data stream or a sequence of signals representing a computer program for performing one of the described methods. The data stream or signal sequence may, for example, be configured to move through a data connection, for example, via the Internet.

A further embodiment comprises processing means, for example, a computer or programmable logic device, configured or adapted to perform one of the described methods.

An additional embodiment comprises a computer, a computer program installed thereon for executing one of the described methods.

An additional embodiment according to the invention comprises a device or system configured to move to a receiver (for example, electronically or optically) a computer program for performing one of the described methods. The receiver may, for example, be a computer, mobile device, storage device or similar device. The device or system may, for example, comprise a file server for transmitting a computer program to the receiver.

In some embodiments, a programmable logic device (eg, a user programmable gate array) may be used to perform some or all of the functionality of the described methods. In some embodiments, a user-programmable gate array may cooperate with a microprocessor to perform one of the described methods. In general, the methods are preferably performed using any hardware device.

The above described embodiments are merely illustrative of the principles of the present invention. It is understood that modifications and changes to the described structures and details will be apparent to those skilled in the art. Therefore, it is intended that the invention be limited only by the scope of the claims of the patent in question, and not by the specific details presented by describing and explaining the embodiments in this work.

Claims (47)

1. Decoder for restoring an array (10) of samples from an entropy encoded data stream, configured for entropy decoding of a plurality of entropy slices in an entropy encoded data stream in order to recover various portions (12) of an array of samples associated with entropy slices, respectively, using performing, for each entropy slice, entropy decoding along the corresponding path (14) of entropy encoding using the corresponding probability estimates, adapting the corresponding probability estimates along the corresponding path of entropy coding using the previously decoded part of the corresponding entropy slice, triggering entropy decoding of multiple entropy slices sequentially using the order (16) of entropy slices, and performing, upon entropy decoding of a predetermined entropy slice, entropy decoding of the current part of the predetermined entropy slice based on the corresponding probability estimates of the predetermined entropy slice, which are adapted using the previously decoded part of the predetermined entropy slice, and probability estimates that were used in the entropy decoding the order of the entropy slices of the previous entropy slice in the next hour te spatially adjacent entropy slice, the decoder is configured to save the probability estimates that appear after the entropy decoding of the second, along the corresponding path of entropy coding, a portion block corresponding, in order of entropy slices, to the previous entropy slice, and use the stored probability estimates for initialization, before decoding, the first, along the corresponding trajectory of entropy coding, a portion block corresponding to the corresponding entropy slice. 2. The decoder according to claim 1, in which the various portions are the rows of blocks of the array of samples. 3. The decoder according to claim 1, in which the order of the entropy slices is chosen so that, in the order of the entropy slices, the different portions follow each other in the direction (16), turned relative to the paths (14) of the entropy encoding of the entropy slices, which, in turn, are distributed parallel to each other. 4. The decoder according to claim 1, in which each entropy slice has entropy encoded data therein for the corresponding portion of the array of samples, different portions form rows of blocks of the array of samples, the blocks being regularly arranged in rows and columns so that the portions corresponding to the entropy slices consisted of the same number of blocks and the path of entropy coding indicated parallel to the lines of blocks, and the decoder is configured to perform, for each entropy slice (90), initialization, for the corresponding entropy slice, probability estimates (94), before decoding, the first portion block (12), the corresponding entropy slice along the corresponding entropy coding path (14), probability estimates that manifest themselves after the entropy decoding of the second portion block (50) (12) corresponding, in order (16) of the entropy slices, to the previous entropy slice along the corresponding path of entropy coding. 5. The decoder according to claim 1, in which each entropy slice has entropy-encoded data in it for the corresponding portion of the array of samples, the various portions form the rows of blocks in the array of samples, the blocks being regularly arranged in rows and columns so that the portions corresponding to the entropy slices consisted of the same number of blocks, and the path of entropy coding indicated parallel along the lines of blocks, and the decoder is configured to perform, for each entropy slice, an entropy decoder along the corresponding path of entropy coding and adapting the corresponding probability estimates along the corresponding path of entropy coding in such a way that after the current part of the predetermined entropy slice is entropy decoded based on the corresponding estimates (94) of the probability of the predetermined entropy slice, the corresponding estimates (94) of the probability of the predetermined entropy slice adapt depending on the current part of the predetermined entropy slice and probability estimates which manifest themselves during entropy decoding of spatially neighboring parts of the neighboring entropy slice. 6. The decoder of claim 5, wherein the decoder is configured to adapt the corresponding probability estimates of the predetermined entropy slice after the current portion of the predetermined entropy slice is entropy decoded based on the corresponding probability estimates of the predetermined entropy slice using the first adaptation in depending on the current part of the predefined entropy slice and averaging the result of the first adaptation using probability estimates, which matured used during entropy decoding of spatially neighboring parts of the neighboring entropy slice. 7. The decoder according to claim 4, wherein the decoder is configured to direct entropy decoding of directly following entropy slices in order of entropy slices so as to prevent the distance between the currently decoded portion blocks corresponding directly immediately following one another entropy slices, measured in blocks along the paths of entropy coding, became less than two blocks. 8. The decoder according to claim 4, in which the decoder is configured to direct entropy decoding of directly following entropy slices in order of entropy slices so that the distance between the currently decoded blocks of portions corresponding directly to successive entropy slices, measured in blocks along the paths of entropy coding, remained equal to two blocks. 9. The decoder according to claim 1, in which the entropy slices are divided into fragments, and the decoder contains a deinterleaver for deinterleaving the fragments and is configured to start entropy decoding of the entropy slices in parallel along the paths of entropy encoding even before receiving any of the entropy slices. 10. The decoder according to claim 1, in which the entropy slices are divided into fragments, and the decoder is configured to checking whether or not the current fragment of the first portion of the portion of the predetermined entropy slice corresponds along the path of entropy coding, and if yes, entropy decoding of the current fragment when adapting the corresponding probability estimates and taking into account the state of the corresponding probability estimates that manifest themselves at the end of the entropy decoding of the current fragment, when another fragment is entropically decoded, which corresponds to the second subportion of a portion of a predetermined entropy slice along the path of entropy encoding, and if not, entropy decoding of the current fragment using probability estimates that depend on the probability estimates manifesting themselves at the end of the entropy decoding of the fragment that corresponds to a subportation of a portion of a predetermined entropy slice preceding a subportion corresponding to the current fragment along the entropy encoding path. 11. The decoder according to claim 1, in which the array (10) of samples is the current array of samples of the sequence of arrays of samples and the decoder is configured to entropy decode the current part of the predetermined entropy slice based on the corresponding estimates of the probability of the predetermined entropy slice entropy slice, which are adapted using the previously decoded portion of the predefined entropy slice, probability estimates used in entropy decoding of a spatially adjacent, in order of entropy slices, previous entropy slice in the neighboring part of a spatially neighboring entropy slice, and probability estimates used in decoding a previously decoded frame of an entropy encoded data stream belonging to a different array of samples than the current array of samples. 12. The decoder according to claim 1, in which the entropy encoded data stream is a temporary multi-level data stream. 13. An encoder for encoding an array of (10) samples into an entropy encoded data stream configured for entropy encoding of a plurality of entropy slices into a data stream of an entropy encoder, each entropy slice is associated with a different portion (12) of the array of samples, respectively, using performing, for each entropy slice, entropy coding along the corresponding path (14) of entropy coding using the corresponding probability estimates, adapting the corresponding probability estimates along the corresponding path of entropy coding using the previously encoded part of the corresponding entropy slice, run entropy coding of multiple entropy slices sequentially using the order (16) of entropy slices, and performing, when entropy encoding a predetermined entropy slice, entropy encoding the current part of a predetermined entropy slice, based on the corresponding probability estimates of the predetermined entropy slice, which are adapted using the previously encoded part of the predetermined entropy slice, and on the probability estimates that were used in the entropy slice encoding the entropy slice in order of entropy slices, previous entropy slice in the adjacent part rostranstvenno neighboring entropy slice, the encoder is configured to save probability estimates that appear after the entropy encoding of the second, along the corresponding path of the entropy encoding, a portion block corresponding, in order of entropy slices, to the previous entropy slice, and use the stored probability estimates for initialization, before encoding, the first, along the corresponding trajectory of entropy coding, a portion block corresponding to the corresponding entropy slice. 14. The encoder of claim 13, wherein the entropy encoded data stream is a temporary multi-level data stream. 15. A method of restoring an array (10) of samples from an entropy encoded data stream, comprising the step of: entropy decode a plurality of entropy slices in the data stream of the entropy encoder so as to restore different portions (12) of the array of samples associated with the entropy slices, respectively, using performing, for each entropy slice, entropy decoding along the corresponding path (14) of entropy encoding using the corresponding probability estimates, adapting the corresponding probability estimates along the corresponding path of entropy coding using the previously decoded part of the corresponding entropy slice, triggering entropy decoding of multiple entropy slices sequentially using the order (16) of entropy slices, and performing, upon entropy decoding of a predetermined entropy slice, entropy decoding of the current part of the predetermined entropy slice based on the corresponding probability estimates of the predetermined entropy slice, which are adapted using the previously decoded part of the predetermined entropy slice, and on the probability estimates that were used in the entropy decoding in order of entropy slices, previous entropy slice in neighboring parts of a spatially adjacent entropy slice, at the same time, the probability estimates that manifest themselves after the entropy decoding of the second, along the corresponding path of the entropy coding, of the portion block corresponding in the order of the entropy slices to the previous entropy slice are stored, and the stored probability estimates are used to initialize, before decoding, the first along the corresponding path of the entropy encoding, a portion block corresponding to the corresponding entropy slice. 16. A method of encoding an array (10) of samples into an entropy encoded data stream, comprising the step of: entropy encode many entropy slices into an entropy encoded data stream, each entropy slice associated with a different portion (12) of the array of samples, respectively, using performing, for each entropy slice, entropy coding along the corresponding path (14) of entropy coding using the corresponding probability estimates, adapting the corresponding probability estimates along the corresponding path of entropy coding using the previously encoded part of the corresponding entropy slice, run entropy coding of multiple entropy slices sequentially using the order (16) of entropy slices, and performing, when entropy encoding a predetermined entropy slice, entropy encoding the current part of a predetermined entropy slice, based on the corresponding probability estimates of the predetermined entropy slice, which are adapted using the previously encoded part of the predetermined entropy slice, and on the probability estimates that were used in the entropy slice encoding the entropy slice in order of entropy slices, previous entropy slice in the adjacent part rostranstvenno neighboring entropy slice, at the same time, the probability estimates that manifest themselves after the entropy encoding of the second, along the corresponding path of the entropy coding, of the portion block corresponding, in the order of entropy slices, to the previous entropy slice are stored, and the stored probability estimates are used to initialize, before encoding, the first, along the corresponding path of the entropy encoding, a portion block corresponding to the corresponding entropy slice. 17. A computer-readable medium storing a computer program having a program code configured to execute, when executed on a computer, the method of claim 15. 18. A computer-readable medium storing a computer program having a program code configured to execute, when executed on a computer, the method of claim 16. 19. A digital storage medium storing an entropy encoded data stream generated by the method of claim 16. 20. The digital storage medium according to claim 19, in which the entropy encoded data stream is a temporary multi-level data stream.

Images